Spatial Patterns of Offender Groups

Abstract

A co-offending network, the network of offenders who have committed crimes together, is a prime source for crime investigation. Analyzing co-offending networks contributes to crime reduction and prevention strategies and tactics at different levels by extracting meaningful patterns and relationships. On a different track, spatial analysis of crime has recently enriched understanding of criminal activity extensively. This study integrates spatial and social network analysis to understand the role of spatial distance in forming criminal collaborations. First, we extract co-offending networks from a police-reported database and present a comprehensive study of the spatial properties of co-offending networks. Then, using community detection approaches, we detect offender groups as denser sub graphs of some co-offending network. Finally, we study the geography of offender groups as an important characteristic of such groups. Recognizing if offender groups are geographically dispersed or geographically concentrated can help law enforcement and intelligence agencies to prioritize their preventative deployments and proactive investigations in combating crime. For the experimental evaluation, we use a real-world crime dataset comprising crime incidents in the time period 2001–2006 in the regions of British Columbia, Canada policed by the RCMP.